This work presents advanced numerical simulation techniques that were developed to improve the history matching and predictive capability of both laboratory and field scale SVX processes. An algorithm was developed to incorporate the time dependence of SVX processes in the CMG STARS software; it incorporated a non-equilibrium solvent solubility method. A new methodology was developed to correct for excessive numerical dispersion effects at field scale gridblock sizes. Thermal effects due to solvent dissolution in heavy oil were also studied.
The advanced numerical simulation technique was constrained by PVT mixing experiments designed to generate mass transfer and non-equilibrium solvent solubility datasets for the solvent–oil system. The numerical models were tuned from the datasets obtained from three lab scale 3D SVX physical model experiments and by the use of a unique set of gas-liquid relative permeability curves (applicable to all three experiments) that employed the solvent mass transfer rate as the history matching parameter. The three physical model experiments varied in only model geometry and horizontal well placement. A relationship between reaction frequency factor (ra) and gridblock size was developed to correct for excessive numerical dispersion effects at field scale gridblock sizes. Thermal effects were incorporated in the numerical model by properly assigning the thermal parameters describing the fluid and porous media heat capacities and vaporization enthalpy coefficients.
The studies found that simulations conducted using a non-equilibrium solvent solubility method yielded more realistic relative permeability (kr) curves. The dissolved solvent concentrations and diluted oil viscosity profiles were also more realistic. The viscosity-reducing potential of the thermal effects of solvent dissolution were found to be negligible when compared with the viscosity-reducing effects of solvent dilution.